\begin{figure}
  \centering
  \subcaptionbox{True counts\label{tab:statistical-true}}{%
    \begin{tabular}{@{}lr@{}}
      \toprule
      Location & \multicolumn{1}{c@{}}{Count} \\
      \midrule
      Belleville & $1$ \\
      Quartier Latin & $1$ \\
      Le Marais & $1$ \\
      Montmartre & $2$ \\
      Opéra & $1$ \\
      \bottomrule
    \end{tabular}%
  }\quad
  \subcaptionbox*{}{%
    \begin{tabular}{@{}c@{}}
      \\ \\ \\
      $\xrightarrow[]{\text{Noise}}$
      \\ \\ \\
    \end{tabular}%
  }\quad
  \subcaptionbox{Perturbed counts\label{tab:statistical-noisy}}{%
    \begin{tabular}{@{}lr@{}}
      \toprule
      Location & \multicolumn{1}{c@{}}{Count} \\
      \midrule
      Belleville & $1$ \\
      Quartier Latin & $0$ \\
      Le Marais & $2$ \\
      Montmartre & $3$ \\
      Opéra & $1$ \\
      \bottomrule
    \end{tabular}%
  }%
  \caption{(a)~The original version of the data of Table~\ref{tab:continuous-statistical}, and (b)~their $1$-differentially event-level private version.}
  \label{fig:scenario-statistical}
\end{figure}